A Federal Twin Platform for Structural Health Monitoring Using the Example of Railway

The ageing infrastructure poses significant challenges for maintaining safety, reliability, and efficiency in transportation networks. With many structures surpassing their intended service life, effective monitoring and maintenance strategies have become imperative. This thesis addresses these challenges by exploring the integration of structural health monitoring and digital twin technologies, presenting a conceptual and technical framework for a centralized data and analysis platform tailored to railway infrastructure, with a focus on bridges. The study begins with an in-depth examination of the current state of monitoring practices, including vision-based and vibration-based monitoring techniques and analysis possibilities in form of physics-based and data-driven approaches. These methods are widely employed to assess structural integrity, yet their adoption is hindered by fragmented data sources and the absence of comprehensive systems for combining diverse information types. Complementing this is an exploration of digital twin concepts, which enable the creation of virtual replicas of physical assets. The thesis distinguishes between digital twins and digital shadows, while outlining the requirements for building robust digital twin networks that support real-time monitoring and predictive maintenance. Building on these foundations, this thesis proposes a novel conceptual architecture for a data and analysis platform, the Federal Twin Platform. The platform centralizes data acquisition, ingestion, storage, and analysis, integrating diverse sources such as sensor measurements, images, inspection reports, and digital models. It connects domain-specific digital twins for key components, including bridges, tracks, trains, stations, and entire railway lines, creating a digital twin network that enhances a holistic decision-making and lifecycle management. Employing a microservice architecture, the platform achieves modularity, scalability, and seamless integration with external systems while supporting both data-driven models and physics-based simulations. The technical implementation of the platform is detailed across several layers, including full-stack development, data pipelines, storage solutions, and visualization services. Key contributions include algorithms for automating data processing, such as location-matching uploaded images and sensor data with GPS data and synchronizing sensor streams with structural models. Machine learning and computer vision techniques are applied to tasks like novelty detection, while advanced signal processing methods are utilized to extract meaningful insights from vibration data. This thesis demonstrates that integrating structural health monitoring and digital twin technologies through a centralized platform can transform how ageing infrastructure is monitored and managed. The proposed, implemented, and deployed framework enhances data integration and analysis, and also serves as a foundation for an intelligent and predictive maintenance system. The core innovation of the Federal Twin Platform is its ability to create a network of digital twins that captures the interdependencies between railway assets and enables a holistic view of the infrastructure. Its evaluation using real-world applications leads to a proof of concept, demonstrating its practical viability and effectiveness. The findings contribute to the advancement of the state of the art in infrastructure monitoring and provide practical insights for the design and implementation of future railway management systems.